The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its core functionality of generating new instruction files (SKILL.md) based on user-provided metadata.
Ingestion points: User-supplied skill name and description are used to populate the content of new SKILL.md files.
Boundary markers: The documentation does not specify the use of delimiters or 'ignore' instructions within the generated SKILL.md to prevent user input from being interpreted as system overrides in the new skill context.
Capability inventory: The skill performs local file system write operations to create directories and files, and it utilizes external Slack notifications via Rube integrations.
Sanitization: There is no evidence of input validation or sanitization to ensure that user-provided descriptions do not contain malicious instructions.
[DYNAMIC_EXECUTION]: The skill programmatically generates directory structures and script files at runtime.
Evidence: The skill is designed to 'Generate standardized scripts/, references/, and assets/ directories' as part of the skill creation process. While this is a standard scaffolding feature, the creation of executable or interpretable content based on external input represents a dynamic execution pattern.

skill-share